DevOpsLinks

Between AI everywhere (in the interview loop, in security workflows, and even in how teams ship code) and a surprisingly spicy systemd change, the vibe is: the tooling is getting powerful and the tradeoffs are getting real. If you have strong opinions about infra ergonomics or where engineering teams are headed next, there is plenty here to sharpen them.

🛠️ 5 Suggestions to Upgrade your OpenTofu/Terraform & AWS Development Experience
🔐 How I Use LLMs for Security Work
🧑‍💻 Software engineer interviews for the age of AI
🗂️ Systemd Gets a birthDate Field and a Liberated Fork in Response
🏭 The Software Factory: Why Your Team Will Never Work the Same Again
🧠 Why system architects now default to Arm in AI data centers

Ship smarter, argue better, and keep your tools on a short leash.

See you in the next issue!
FAUN.dev() Team

systemd merged a PR that adds an optional JSON user record field to userdb. Only a sysadmin can write it. The field targets California, Colorado, and Brazil age laws. A revert PR was rejected. The solo fork Liberated systemd strips the addition.

The current models and tooling are enough to build software factories. In a software factory, developers stop writing code by hand, and AI coding agents implement features and fix bugs while developers design and improve the factory. Tools like Claude Code and Gas Town enable this shift towards a more efficient and streamlined production workflow.

System shift: Today, implementations are pivoting to agent-driven factories. Teams are shrinking. Investment is narrowing to skills, architecture, and CI backpressure.

The article covers tools and scripts to reclaim focus and improve workflow for OpenTofu, Terraform, and AWS CLI users. Suggestions include tools for easily swapping between versions, summarizing plans, linting code, switching AWS profiles, and customizing prompts. Bonus recommendation includes Task for automating bulk operations across multiple projects.

AI is becoming more prevalent in coding interviews, sparking interest from experienced candidates tired of traditional methods. Hiring great engineers is crucial for maintaining reliable services, especially in the era of AI-generated code. System design interviews help identify candidates with hands-on experience and the ability to navigate unknown codebases.

The writer shared their experience of using LLM tools like Claude, Cursor, and ChatGPT in security and engineering work. They emphasized the importance of specific and role-stacking prompts when using these tools, and highlighted the need to provide context and constraints for more accurate responses. Additionally, they discussed techniques for optimizing the use of LLMs in security work, and encouraged sharing effective prompting patterns within the community.

AI is reshaping infrastructure needs, exposing the limits of legacy architectures. As AI workloads demand more efficient platforms, design is shifting towards purpose-built rack-level systems. Arm-based architectures are being chosen to address constraints shaping modern AI platforms, with a focus on system-level harmony and CPU performance.

The open-source server command center that replaces Portainer + pgAdmin + Uptime Kuma + your deployment scripts — with 4 npm dependencies.

AI-Powered Autonomous SRE Platform — Autonomous agents for investigation, remediation, security, and learning across multi-cloud and on-prem infrastructure

Let agents test your code in a real browser

Terravision creates Professional Cloud Architecture Diagrams from your Terraform code automatically. Supports AWS, Google and Azure.

Did you know that systemd has a built-in software watchdog that goes beyond simple crash detection? With Type=notify and WatchdogSec=, a service must send periodic WATCHDOG=1 keep-alive pings (via sd_notify) at roughly half the configured interval - if systemd stops receiving them, it kills and restarts the service, catching hung processes that are alive as PIDs but stuck in a deadlock or infinite loop. The same Type=notify mechanism also requires the service to explicitly send READY=1 before systemd considers it started, so systemctl start blocks until the process has actually bound its sockets - eliminating the "started but not ready" race. Paired with Restart=on-failure and cgroup-backed directives like MemoryMax= and CPUQuota=, this gives each service unit independent, tight failure semantics without any external supervisor.

"Every guardrail you add in the cloud buys reliability by selling understanding; soon the system is stable and nobody knows how it works. Then the first real incident isn't a failure of servers, it's a failure of assumptions."
— SenseiOne

Trusting systemd unit restart logs as your primary signal for service health means you're measuring recovery speed, not failure frequency - a service that restarts cleanly every 90 seconds looks stable in uptime dashboards while the underlying segfault or resource exhaustion goes uninvestigated for weeks.

This week, we’re highlighting Dirceu Vieira Junior, a Senior Software Engineer at iFood and 7x Salesforce Certified full stack developer with 10+ years in software. Formerly a Salesforce Tech Lead at Atrium, he has spent the last decade building Salesforce systems end to end.